SPOTLIGHT ON GENETICS

Making waves in genomics

Building personalised medicine is as much about images as it is about omics. By combining the two, we can really move towards precision medicine for individuals. Gabriel Krestin, Erasmus University Medical Centre in Rotterdam, the Netherlands

The big data of omics research and advanced imaging technology are being combined to understand, diagnose and treat disease.

THE LONG-ACCEPTED narrative that radiation merely kills cells and damages DNA is officially a thing of the past, according to Gillian Tozer, from the Department of Oncology at Sheffield University in the UK. There has been much progress in radiation biology since X-rays were first used to treat cancer a few years after their discovery in 1895. After decades of steady progress, radiation development has reached an accelerating phase as the rapid accumulation of molecular and genomic information has given birth to the field of radiogenomics, which means different things in varying contexts.

Cancer radiogenomics

Irradiated mouse tumour with damaged cells (blue) and disrupted layers of endothelial cells (brown) that line all blood vessels. Credit: GILLIAN TOZER

Coined in the early 2000s, the term was initially used by cancer physicians to describe the use of features of a patient's genome to predict their responses to radiation therapy across many types of cancer. More than 50% of cancer patients receive radiation therapy at some stage of their treatment, and the first focus of associated genomics research was to identify common genetic variations, such as single nucleotide polymorphisms (SNPs), that influence a patient's likelihood of developing toxicity to the therapy.

But, scientists are learning of the effects of such treatment beyond toxicity. Radiation can induce gene expression, alter signalling pathways and affect the blood supply to a tissue. This effect of radiation on a tumour's vasculature is Tozer's particular expertise. “A key question in the field is how radiotherapy can best be combined with specific drugs, such as those developed to inhibit blood vessel growth or function,” she says.

Tozer says that this broadening of scope has brought radiation-biology research out of a slump. “The funding situation has been up and down over the years, but it's picked up in the past decade or so.”

The Medical Research Council (MRC) in the UK, for example, is particularly encouraging grant applications in radiation oncology and biology. This is opening opportunities for new and existing researchers, such as at the Oxford Institute for Radiation Oncology and Biology, a joint initiative of Cancer Research UK and the MRC, which opened in 2008.

In 2009, the UK's National Cancer Research Institute (NCRI) also set up CTRad, a working group of clinicians and scientists focused on enhancing radiation research in the UK and for developing radiotherapy-related clinical trials.

In the USA, the Radiogenomics Consortium was established in 2009 and now involves more than 150 researchers across 19 countries. The main aim of the group is to identify common genetic variations, associated with radiation toxicity. The consortium also facilitates international collaborations by sharing data and specimens, develops standards for radiogenomics research and encourages meta-analyses of relevant studies.

The resurgence of interest in radiation biology is leading to immediate improvements for patients as well as enhanced understanding of basic biology. In particular, using real-time imaging during radiation therapy allows more accurate targeting of the radiation and determination of changes in tumour size or exact position of an organ.

Information and imaging

The newer and more widely used interpretation of the term radiogenomics is the interface of imaging with radiation biology: linking information from the booming ‘omics’ sciences — genomics, transcriptomics and proteomics — with medical imaging in all its forms, from magnetic resonance imaging (MRI) through to X-ray computerised tomography (CT) scans. The primary aim is to identify image characteristics, or biomarkers, that can be used to diagnose and treat conditions including neurodegeneration, cardiovascular disease, arthritis and cancer. It is hoped that such biomarkers will reduce the necessity for tissue biopsies, allowing more rapid and frequent monitoring of a disease.

To identify image biomarkers, researchers look for associations between image features — such as tissue morphology, texture or density — and molecular or genetic characteristics, for example the expression of proteins that indicate a specific tumour type or that are predictive of a patient's prognosis or response to a treatment. Gabriel Krestin from the Erasmus University Medical Centre in Rotterdam in the Netherlands says that scientists have long believed there was much more information to be gleaned from images than what meets the eye, and the underlying biological reasons for particular tissue characteristics are starting to become clear. For instance, we now know that certain image features can be associated with a tumour's aggression or can point to tissue hypoxia, a frequent tumour characteristic that not only alters responses to radiation and other therapies, but may also promote metastasis of the cancer cells.

Making these associations is not straightforward. “We can get 120 or more imaging descriptions of a tumour,” says Krestin. The large volume of information is incredibly useful, but it turns what was once a highly qualitative science into a highly quantitative one. Scientists who are familiar with ‘big data’ are stepping in to help.

You need at least a desire to be able to understand some of the language of the other scientists you'll be working with Gillian Tozer, Sheffield University

One such researcher is Olivier Gevaert at Stanford University in California. The Belgian started his career at the KU Leuven, using Bayesian networks to model omics data and predict diagnosis, prognosis or therapy responses in cancer patients. In 2009 he moved to Stanford to work in their radiology department as part of their Integrative Cancer Biology Program. “I started linking their imaging expertise with my work on omics data,” he says. Now, he describes his lab's work as ‘multiscale data fusion’. “We consider a tumour on a molecular level, looking at gene expression, copy number, epigenetic patterns. We look at it on a cellular level, using pathology techniques such as staining, and we look at it on a tissue level, using medical imaging. And then we try to integrate all of that information.”

Radiogenomics research is highly collaborative, involving scientists with backgrounds in radiology, molecular biology, genomics, informatics and imaging analysis. It has taken a while for this expertise to start to come together, and Gevaert estimates that the field is five to ten years behind genomics in certain respects. For example, journals often don't require imaging data to be published in the same way that genomics data are.

“We used to have very few datasets with both imaging and molecular data from the same tumour sample,” says Gevaert. But momentum is building and leading to increased awareness of the value of such matched samples. A catalyst for the field has been The Cancer Imaging Archive (TCIA), an ever-growing collection of cancer-related medical images to which researchers around the world have free access. A main goal of the archive, which is curated by the Cancer Imaging Program of the US National Cancer Institute (NCI), is to gather clinical diagnostic images that match patient cases for which there is genomic information available as part of The Cancer Genome Atlas. The TCGA project, initiated in 2006 and sponsored by NCI and the National Human Genome Research Institute, is collating genomic information on more than 20 cancer types from source institutions across the US, and these data are also freely available.

It is hoped that generating large databases of data from many individuals will eventually lead to care which is much more patient specific. “Building personalised medicine is as much about images as it is about omics,” says Krestin. “By combining the two, we can really move towards precision medicine for individuals.”

Although the main focus of imaging genomics research has been in cancer thus far, researchers are using the same concepts to study other diseases. For example, combining knowledge of medical images and genetic information might help predict an individual's risk of developing Alzheimer's disease. And right at the cutting edge of the field of radiogenomics is the idea that radiation might be used to modulate gene expression (see Natural nanoparticles).

Researchers interested in this fast-moving field should approach it with an open mind. “You need at least a desire to be able to understand some of the language of the other scientists you'll be working with,” says Tozer. Somehow, the statistics and cell biology of omics, the physics of radiation biology and imaging, and the clinician's experience of treating individual patients need to come together. But radiogenomics is a relatively young field and it has had a highly collaborative start. “People talk about fields becoming interdisciplinary, but we always have been,” says Krestin. “You could say it's in our genes.”

Natural nanoparticles

The technique of optogenetics, in which light-sensitive proteins are used to regulate the activity of specific neurons, has been a game-changer in neuroscience research. But the process requires a light source to be implanted into the brain of experimental animals. Jeffrey Friedman at Rockefeller University in New York and his colleague Sarah Stanley, now at Mount Sinai Hospital, are trying to accomplish the same goal in a non-invasive manner. His lab is pioneering a technique that relies on natural nanoparticles — specifically, ferritin, protein particles that store iron within cells. These can heat up in response to low-frequency radio waves.

Jeffrey Friedman and Sarah Stanley. Credit: ZACK VEILLEUX

The group have created a modified gene that encodes ferritin expressed close to a heat-activated ion channel in the cell membrane. Opening this channel causes calcium ions to move into the cell. The researchers then added a gene that expresses insulin in a calcium-dependent manner. By expressing this combined gene construct in mice, using either stem cells or a lentivirus, and then exposing the mice to radio waves or a magnetic field, they could induce the ferritin nanoparticles to move such that the ion channel opened and insulin expression was switched on.

This proof-of-principle experiment was published in Nature Medicine in December 2014. Friedman says he's excited about adapting the approach to other scenarios. “We're now exploring the utility of the method to control neural activity,” he says. Much further down the track, the technique might be used for long-term modulation of neural activity in humans, such as to treat severe metabolic disease or Parkinson's. But alongside all the biological challenges is a more technical hurdle. “For the mice experiments, we use a commercial welding apparatus to create the radio waves,” says Friedman. This is where engineers join the radiogenomics field. Several stakeholders, including personnel from Google, are already developing small and safe devices to generate radio waves for medical use.